INDEX

Explanations

words related to being physically light or weight

oai_token-act-pair · gpt-3.5-turbo

references to the terms "lighter" and "heavier," often in relation to various objects or contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

raltar

-0.88

IDENT

-0.83

posted

-0.81

advertising

-0.78

AIDS

-0.78

EP

-0.78

event

-0.78

WAR

-0.76

HI

-0.76

ULE

-0.73

POSITIVE LOGITS

 lighter

1.13

 than

1.02

 weights

0.95

iating

0.90

 Than

0.90

weights

0.88

 "$:/

0.88

weight

0.86

 weight

0.85

 bulb

0.84

Activations Density 0.006%

words related to being physically light or weight

references to the terms "lighter" and "heavier," often in relation to various objects or contexts

No Comments

No Known Activations